Dario reported in the tag up on 9/25 that the RAs and decs of sources in the L2 and L3 prompt catalogs were wildly incorrect, likely owing to using a stale WCS. Check to see if this is still the case. His report is in SV3.pdf linked in the 9/25 agenda minutes.

Roughly this would mean:

• generate an L1 image anywhere except at 250 +50 using romanisim using a Gaia ecsv catalog
• run romancal on the L1 image to produce an L2 image and catalog
• compare the RA and DEC of the input and output catalogs and look for a large offset (degrees)
• run the mosaic pipeline on the L2 product to create an L3 product and catalog
• compare the RA and DEC of the input and output catalogs and look for a large offset

@stscijgbot-rstdms